Identification of Crystalline Orientation of Tungsten Tips by Machine Learning Analysis of Field Ion Micrographs

Abstract

We proposed an automatic procedure of identification of crystalline orientation of W tips observed by field ion microscopy by means of machine-learning-based objective detection and classification. The identification is performed with the geometrical position of {110} and {112} planes detected by YOLOv3, and the correct crystalline orientation was output for 85% of the total data by k-nearest neighbor algorithm, indicating the effectiveness of this method.